Contribution of Extreme Convective Storms to Rainfall in South America

K. L. RASMUSSEN,* M. M. CHAPLIN, M. D. ZULUAGA,1 AND R. A. HOUZE JR.

Department of Atmospheric Sciences, University of Washington, Seattle, Washington

(Manuscript received 24 April 2015, in final form 14 August 2015)

ABSTRACT

The contribution of extreme convective storms to rainfall in SouthAmerica is investigated using 15 years of

high-resolution data from the Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR).

Precipitation from three specific types of storms with extreme horizontal and vertical dimensions have been

calculated and compared to the climatological rain. The tropical and subtropical regions of South America

differ markedly in the influence of storms with extreme dimensions. The tropical regions, especially the

Amazon basin, have aspects similar to oceanic convection. Convection in the subtropical regions, centered on

La Plata basin, exhibits patterns consistent with storm life cycles initiating in the foothills of the Andes and

growing into larger mesoscale convective systems that propagate to the east. In La Plata basin, convective

storms with a large horizontal dimension contribute ;44% of the rain and the accumulated influence of all

three types of storms with extreme characteristics produce ;95% of the total precipitation in the

austral summer.

1. Introduction

Precipitation from thunderstorms and mesoscale

convective systems (MCSs) greatly influences agricul-

tural and socioeconomic conditions in South America.

Yet, these storms often occur in regions without routine

ground-based meteorological observations. The launch

of satellites with spaceborne radars has made it possible

to study the physical characteristics of storms in such

regions, and the Tropical Rainfall Measuring Mission

(TRMM) satellite with its Precipitation Radar (PR) was

in orbit long enough to evaluate these characteristics

climatologically. The satellite radar’s ability to discern

storm structures in three dimensionsmakes it possible to

determine the nature of the precipitating systems pro-

ducing the rainfall. Using these radar data to study the

frequency, intensity, and structure of extreme precipita-

tion events in the current climate will lay groundwork for

anticipating changes in these patterns of precipitating

convection as climate changes occur.

It is important to examine the climatology of storm

type as well as rainfall. In general, regions experiencing

the most precipitation typically do not coincide with

regions known to support storms with the most extreme

vertical structure (Zipser et al. 2006). Hence, flash

flooding and severe local weather may be favored de-

spite low overall rainfall. On the other hand, MCSs

(Houze 2004), characterized as storms with large hori-

zontal dimensions, can contribute large fractions of warm

season rainfall because of their breadth, long-lived na-

ture, and repeated occurrence in certain regions (Fritsch

et al. 1986; Durkee et al. 2009). It therefore seems im-

portant to determine the climatology of convective

storms with differing horizontal and vertical scales.

The goal of this study is to assess the relative contri-

bution of precipitation from storms with the most ex-

treme horizontal and vertical dimensions to the

climatological precipitation in South America. Some

past studies have focused on the contributions to rainfall

by systems of different horizontal dimension (e.g., Liu

2011; Romatschke and Houze 2013). Other studies have

examined the statistics of vertical echo structure (e.g.,

Boccippio et al. 2005). In the present study, we use a

technique that uses metrics of both vertical and hori-

zontal dimensions to identify deep convective systems of

different types, corresponding to different stages of

* Current affiliation: National Center for Atmospheric Research,

Boulder, Colorado.1Current affiliation:UniversidadNacional deColombia,Medellín,

Colombia.

Corresponding author address: Kristen Lani Rasmussen, Na-

tional Center for Atmospheric Research, 3450 Mitchell Lane,

Boulder, CO 80301.

E-mail: kristenr@ucar.edu

JANUARY 2016 RASMUS SEN ET AL . 353

DOI: 10.1175/JHM-D-15-0067.1

� 2016 American Meteorological Society

convective system development. Romatschke andHouze

(2010), Rasmussen and Houze (2011), and Rasmussen

et al. (2014) have used this methodology to examine cli-

matological patterns of convection across South Amer-

ica. These studies showed how the storms vary regionally

in character. However, these studies did not determine

the contribution from storms of different types and in

different stages of development to the rainfall climatol-

ogy of the continent. That is the purpose of this paper.

2. Data and methodology

a. Regions of study

For reference, Fig. 1 shows the regions examined in

this study as well as the major rivers in South America.

Our methodology identifies the contributions by dif-

ferent storm types in the indicated regions.

b. Identification of extreme echo features and stormtypes

The TRMM PR has fine three-dimensional spatial

resolution (4–5 km horizontal, 250m vertical) with a

near-uniform quasi-global coverage that permits com-

prehensive analysis between 368Nand 368S (Kummerow

et al. 1998, 2000). This study uses 15 years of TRMMPR,

version 7 (V7), data from January 1998 to December

2012 for the austral spring through fall seasons.We focus

on the austral spring [September–November (SON)]

and summer [December–February (DJF)] to specifically

address the impact of storms with extreme characteris-

tics on warm season convective precipitation. Addi-

tionally, Houze et al. (2015) demonstrate that in regions

experiencing frequent midlatitude frontal systems dur-

ing the winter, stratiform echoes are more likely to be

produced by frontogenetical processes rather than hav-

ing their origin in convective processes. Thus, we limit

the focus of this study to the seasons most likely to ex-

perience convective storms and MCSs in South Amer-

ica. The following TRMM data products are used:

d 2A23 (rain characteristics; Awaka et al. 1997), where

rain is separated into three categories: convective,

stratiform, and other, and all references to convective

and stratiform precipitation are based on these classi-

fications; andd 2A25 (rainfall rate and profile; Iguchi et al. 2000), which

provides the attenuation-corrected three-dimensional

reflectivity data.

These data were processed following the methodology

of Houze et al. (2007) and Romatschke et al. (2010). All

of the fields were mapped onto a 0.058 3 0.058 latitude–longitude Cartesian grid. We first identify all three-

dimensional echo objects that have detectable radar

reflectivity consisting of two or more contiguous hori-

zontal pixels. Each such object is defined as a distin-

guishable rain area (DRA; Houze et al. 2015).

All pixels located in DRAs are identified, and the

precipitation rates in those pixels are calculated. We

search each DRA to determine if it contains embedded

within it certain extreme characteristics. The embedded

features that we identify here have been defined and

used in previous studies of continental convection by

Houze et al. (2007), Romatschke and Houze (2010),

Romatschke et al. (2010), Houze et al. (2011), Rasmussen

and Houze (2011), Rasmussen et al. (2013, 2014, 2015),

Zuluaga andHouze (2015), Rasmussen andHouze (2015,

manuscript submitted to Mon. Wea. Rev.), and Houze

et al. (2015). These embedded features are defined as

1) deep convective core (DCC), which is a three-

dimensional contiguous 40-dBZ echo $10 km in maxi-

mum height; 2) wide convective core (WCC), which is a

three-dimensional contiguous 40-dBZ echo with a

maximum horizontal dimension $1000 km2; and 3)

broad stratiform region (BSR), which is a region of

contiguous stratiform echo $50 000km2 in horizontal

dimension. These embedded echo features have a re-

lationship to the stage of development of the storm

producing the echoes. DCCs represent especially in-

tense convection and tend to be found in earlier stages

of development. WCCs represent very intense con-

vection that has grown upscale to form amesoscale unit

of intense convection. BSRs occur in the mature and

later stages of development of MCSs (Houze 2004).

Using ground-based radar, Zuluaga and Houze (2013)

demonstrated that these types of echo features indeed

FIG. 1. Topographical map of SouthAmerica showing theAndes

mountain range, associated terrain features, and major rivers. Se-

lected regions for this study are outlined in maroon and labeled.

354 JOURNAL OF HYDROMETEOROLOGY VOLUME 17

represent early, middle, and later stages of convective

system development in a statistical sense.

We define ‘‘storm type’’ according to whether a DRA

contains one of the categories of embedded extreme

echo features (DCC, WCC, or BSR). If so, then the

DRA is referred to as a storm containing a DCC, WCC,

or BSR. Then, we determine whether or not the rain

observed by the TRMMPR at a given location and time

was falling from a DRA containing one of the above

echo types. Themain objective of the paper is to compile

rainfall statistics on storms containing DCCs, WCCs,

and BSRs. We have also defined a separate category of

storms containing both DCCs and WCCs. However, for

conciseness we have not included this combined cate-

gory in some of the maps presented below. Suffice it to

say that those mapped patterns lie between those of

storms containing DCCs andWCCs, and we will present

summary statistics that include this combined category.

c. Precipitation estimation

Iguchi et al. (2009) suggested that the TRMM PR V7

2A25 rainfall algorithm tends to underestimate the

precipitation from deep convection over land.

Rasmussen et al. (2013) investigated the scope of this

bias in extreme storms in South America and confirmed

that the TRMM PR V7 2A25 algorithm tends to un-

derestimate rain in all three extreme echo types used in

the current study because of insufficiencies in the rain

algorithm in capturing the full characteristics of deep

convective storms over land regions. Rasmussen et al.

(2013) showed that lower estimates by the algorithm are

most biased for extreme precipitating systems that

contain significant mixed phase and/or frozen hydro-

meteors. Regions of South America that experience the

most frequent storms containing deep convective cores

are in the subtropics that do not regularly receive large

amounts of climatological rainfall. Thus, an underestima-

tion of the climatological precipitation can influence the

perception of the climatology and hydrologic cycle in South

America.

To mitigate the TRMM PR algorithm bias for the

types of overland storms studied here, we adopted the

methodology of Rasmussen et al. (2013), who proposed

using the Z–R relationship

Z5 aRb , (1)

where Z is the equivalent radar reflectivity factor

(mm6m23) and R is the rain rate (mmh21). This re-

lationship is used to estimate surface precipitation from

the TRMM PR attenuation-corrected reflectivity data.

The lowest nonzero value of Z is used at each data pixel

between the surface and 2.5 km above ground level for

each precipitation echo, which is similar to the procedure

used for the TRMM PR algorithm 2A25 (Rasmussen

et al. 2013). The parameters a and b are constants de-

pending on rain type (convective, stratiform, or other).

Rasmussen et al. (2013) examined multiple values for

these parameters. Here we use values used previously by

Romatschke and Houze (2011), which give a reasonable

estimate of precipitation in the tropics and subtropics

(convective: a 5 100, b 5 1.7; stratiform: a 5 200, b 51.49; and other: a 5 140, b 5 1.6).

d. Calculation of storm type rain contribution

One complication in using the TRMM PR data to

develop statistics of rainfall by certain storm types is that

the sampling by satellite overpasses is intermittent in

time and space (Negri et al. 2002). For statistics to be

comparable between locations, we must make an ac-

commodation for the irregular sampling. To make the

data comparable, we define a parameter F to be applied

to every horizontal grid element (pixel):

F5

�N

S

NT

�, (2)

where NT is the total number of times a given pixel is

sampled by a TRMM overpass and NS is the number of

times that pixel is occupied by a certain storm type

(defined in section 2b) at the time of an overpass. The

contribution C of a given storm type to the rainfall at

that pixel is then calculated according to

C5F

R

S

RT

!, (3)

where RT is the average rain rate (mmh21) seen in a

given pixel over the 15 years of TRMM measurements

and RS is the average rain rate within the subset of those

grid elements that are occupied by a given storm type.

The resulting values represent a satellite overpass-

corrected field of the rain contribution from each

storm type to the total precipitation in the study region.

By using this technique, this study will assess how much

of the climatological rain is contributed by each extreme

storm type and provide insights into the influence of

extreme-storm-related precipitation on the hydrologic

cycle in various regions of South America.

3. Background climatology of precipitation andradar echo characteristics

a. South American hydroclimate

The hydroclimate of South America varies strongly

from the tropical to subtropical regions. Tropical South

JANUARY 2016 RASMUS SEN ET AL . 355

America, largely characterized by the Amazon River

basin, has a pronounced annual cycle of precipitation

that supports the largest rain forest in the world and

contributes 20% of the global river freshwater discharge

to oceans (Richey et al. 1989). Seasonal variations of the

intertropical convergence zone (ITCZ) and South At-

lantic convergence zone (SACZ) result in wet and dry

seasons in tropical South America in the summer and

winter, respectively (Kodama 1992; Nogués-Paegle and

Mo 1997; Carvalho et al. 2004). In contrast, subtropical

South America, largely characterized by La Plata River

basin, receives much less precipitation than the tropical

regions in general, but the precipitation variability in the

spring and summer is related to the occurrence of both

deep convection and MCSs in this region (Velasco and

Fritsch 1987; Zipser et al. 2006; Salio et al. 2007;

Romatschke and Houze 2011; Rasmussen and Houze

2011; Houze et al. 2015).

The Andes mountain range affects the hydroclimate

of both tropical and subtropical South America. As one

of the largest and longest mountain ranges on Earth, the

Andes influence the exchange ofmoisture and heat from

the tropical to subtropical regions primarily through the

South American low-level jet (SALLJ; Vera et al. 2006)

along the eastern foothills of the Andes. Deep convec-

tion in subtropical South America is related to the

presence of the SALLJ (Salio et al. 2007; Rasmussen

and Houze 2011; Rasmussen and Houze 2015, manu-

script submitted to Mon. Wea. Rev.). In addition, the

relationship between steep-sloping terrain and heavy

rainfall is particularly important for the susceptibility to

both flash and slow-rise river flooding and agricultural

sustainability in subtropical South America (Latrubesse

and Brea 2009). This investigation of the hydroclimate

characteristics in tropical and subtropical South Amer-

ica, specifically focused on the most extreme precipita-

tion elements seen by the TRMM satellite, is important

for understanding the hydrometeorology in South

America with implications for forecasting, agriculture,

human consumption, hydropower, streamflow charac-

teristics in tropical versus subtropical river basins, and

future climate projections.

b. TRMM radar echo characteristics

Figure 2 shows the probability of finding a DRA

during the austral spring (i.e., SON; Fig. 2a) and the

climatology of precipitation generated by these events

(Fig. 2b). Although the precipitation maps shown in

Fig. 2b and throughout this study are created with

TRMM PR orbital data (i.e., instantaneous measure-

ments with TRMM PR over an orbital swath), the

spatial patterns of rainfall in these 15-yr-long samples

are in agreement with climatologies of precipitation

relying on continuous and merged multisensor mea-

surements (e.g., Huffman et al. 2001; Rozante et al.

2010; Liu 2015). In the austral spring, more DRAs tend

to occur in the tropical Amazon basin than the sub-

tropics, particularly near the Andes foothills (Fig. 2a).

The precipitation climatology for the spring shows an

especially robust rain maximum in southern Brazil and

northeastern Argentina, likely related to frequent

MCSs in this region. Rasmussen et al. (2014) examined

the lightning and severe storm characteristics of storms

in the austral spring and showed that maxima in light-

ning flash rates and flood storm reports are collocated

with the rain maximum seen in Fig. 2b, likely related to

storms with both deep and wide intense convective

characteristics.

FIG. 2. (a) Geographical distribution of the probability of finding a DRA ($2 pixels) during the austral spring

(SON) from 1998 to 2012. (b) Precipitation climatology for all DRAs identified in (a). The contour inside the

continent represents the 500-m terrain elevation.

356 JOURNAL OF HYDROMETEOROLOGY VOLUME 17

Figure 3 shows similar maps for the austral summer

(i.e., DJF). The contrast between the tropics and sub-

tropics regarding the number of DRAs is much greater

than for the austral spring (Fig. 3a). The Amazon basin

has a large amount of climatological rain contributed by

many raining events in the region. However, the sub-

tropics receive a substantial amount of rainfall despite

the low number of events there compared to the tropics.

Similar to the austral spring (Fig. 2), the subtropical

rainfall is produced by fewer events than tropical rain-

fall. Additionally, the subtropical region exhibits a dis-

tinct longitudinal shift in the probability of DRAs and

the climatological precipitation to the west toward the

Andean foothills. A similar shift in extreme storm oc-

currence and lightning production is shown inRasmussen

et al. (2014), consistent with convective initiation and

upscale growth into MCSs in subtropical South America

(Romatschke and Houze 2010; Rasmussen and Houze

2011; Rasmussen and Houze 2015, manuscript submitted

to Mon. Wea. Rev.).

Figure 4 presents an overview of the probability of

finding the three categories of extreme radar echo

structures with the TRMM satellite (defined in section

2b) for both the spring and summer. DCCs are notably

absent in the tropical Amazon (Figs. 4a,d). WCCs and

BSRs do occur in the Amazon region, though not as

frequently as in the subtropics. Comparing to Figs. 2 and

3 and a recent study examining the variable nature of

convection globally from TRMM PR data (Houze et al.

2015), we conclude that the convective elements in this

region have a less deep maritime-like character, consis-

tent with Mohr et al. (1999), but nevertheless can aggre-

gate into mesoscale units and overall contribute a large

amount of precipitation in the Amazon basin. This char-

acteristic of convection being weaker but nevertheless

productive of rainfall verifies the frequently made claim

that the Amazon region is climatologically similar to a

tropical ocean; sometimes this characterization is referred

to as the Amazon region being a ‘‘green ocean.’’ The

Amazon region has large expanses of open water, a moist

and shallow boundary layer, and limited surface temper-

ature variability. Recent studies have demonstrated that

mesoscale variations in surface heating are an important

factor for convective intensity variability over oceans and

continents (Robinson et al. 2011). Additionally, Wall

et al. (2014) show that the absence of strong low-level

wind shear and convergence in tropical South America

contributes to the maritime character of the Amazon.

Figure 4 is an objective verification of this oceanic char-

acteristic of Amazon convective precipitation. Houze

et al. (2015) show that the lack of DCC echoes, especially

in summer (Fig. 4d), and the frequent occurrence of BSR

echoes (Fig. 4f) are characteristics similar to those seen

over the tropical oceans.

The regions with the highest probability of extreme

echo occurrence in the subtropics are collocated with

the maximum in precipitation in the spring and summer.

Previous studies of subtropical South American convec-

tive systems in the austral summer have hypothesized a

storm life cycle where convection initiates along the

Andean foothills and sometimes grows upscale into

MCSs while propagating east or northeast and then de-

cays into broad stratiform regions farther east (Romatschke

and Houze 2010; Matsudo and Salio 2011; Rasmussen and

Houze 2011). The storm type distributions of DCCs,

WCCs, and BSRs in spring are heavily concentrated in

southern Brazil and northeastern Argentina, unlike the

summer patterns. This difference suggests that a some-

what different canonical storm life cycle occurs during the

austral spring and should be investigated in a future

FIG. 3. As in Fig. 2, but for the austral summer (DJF).

JANUARY 2016 RASMUS SEN ET AL . 357

study.However, as will be discussed below in section 5, an

analysis of the diurnal cycle of extreme-storm-contributed

rainfall in both the spring and summer show similar east-

ward propagation and upscale growth from the late af-

ternoon through early morning, but with decreased

magnitudes in the spring.

Table 1 shows the overall statistics of the total number

of DRAs and TRMM-identified extreme echo cores

(defined in section 2b) for the austral spring through fall

seasons, identified in each region defined by the boxes in

Fig. 1. In general, more storms containing WCCs were

identified in most of the regions. However, the ratio of

FIG. 4. Geographical distribution of the probability of finding an event by each extreme type during the austral

(left) spring (SON) and (right) summer (DJF) from1998 to 2012. The contour inside the continent represents the 500-m

terrain elevation.

358 JOURNAL OF HYDROMETEOROLOGY VOLUME 17

the number of extreme cores to the total DRAs is very

low (Table 2). The highest percentages of extreme

storms to total events are WCCs in the northern and

southern La Plata regions (1.8% and 1.9%, respectively)

that experience frequent MCSs in the spring and sum-

mer. As seen in Table 3 showing the average size of the

embedded extreme echoes in various regions in South

America, WCCs cover a much larger area than DCCs

and thus likely contribute more precipitation. Thus,

;2% of the storms in La Plata basin likely make up a

large fraction of the climatological rain, echoing the

discussion of Figs. 2 and 3 above. This type of assessment

of how much rain each storm type contributes provides

crucial information for water management and extreme-

storm-related flood risks in subtropical South America.

4. Rainfall contributions by storms containingextreme radar echo structures

We now examine the contributions of the pre-

cipitation produced by various types of storms on South

American hydroclimate. A ‘‘storm type’’ is assigned to a

DRA (as defined in section 2b) if it contains an em-

bedded extreme echo of one or more of the types de-

fined in section 2b as DCC, WCC, and BSR. For the

precipitation at a pixel to be attributed to one of these

storm types, the rain falling must be from a DRA that

contains one of these three extreme elements. Figures 5

and 6 represent the spatial distribution of the rainfall

contribution C, defined in (3), to the total rain by storm

types containing DCCs, WCCs, and BSRs during the

austral spring and summer, respectively.

Overall, storms containing WCCs are the strongest

contributors to the climatological rainfall in South

America in both spring and summer (Figs. 5b, 6b).

Storms containing DCCs have a moderately strong

impact in southwestern Argentina (Fig. 6a), but other-

wise contribute relatively little to total rainfall. The

storms with DCCs are smaller in horizontal scale and

shorter lived than storms that containWCCs or BSRs. In

the subtropical zone, the precipitation contribution from

storms containing BSRs tends to maximize just east of

the region where storms containing WCCs are most fre-

quent, consistent with the hypothesis that storms grow

and decay as they propagate eastward (Romatschke and

Houze 2010; Matsudo and Salio 2011; Rasmussen and

Houze 2011). The rain contributions by storms with wide

cores are of extreme importance in the subtropical spring,

most notably over La Plata basin, where the maximum of

WCC occurrence and precipitation are collocated (cf.

Figs. 4b, 5b). Storms containing WCCs have both very

intense rainfall and have organized up to a larger hori-

zontal scale. Thus, they have the double advantage of

heavy rain elements and large area.

Consistent with the southwestward shift from the

spring to summer of extreme echo type probability

(Fig. 4), the rain contributions for each storm type also

shift southwestward toward the Andean foothills during

summer (Fig. 6). Storms with DCCs contribute slightly

more rain to the total compared to spring, and the pat-

tern is focused around the Sierras de Córdoba (small

north–south secondary mountain feature east of the

main Andes barrier between ;308 and 358S) in Fig. 6a,

suggesting stronger orographic forcing in the summer.

Similar to Fig. 5, storms containingWCCs contribute the

most climatological rain in the subtropics near central

Argentina and Uruguay. This region is important to the

hydroclimate of South America because frequent ex-

treme MCSs occur here (Romatschke and Houze 2010;

Rasmussen and Houze 2011; Rasmussen et al. 2014;

Rasmussen and Houze 2015, manuscript submitted to

Mon. Wea. Rev.). Rasmussen et al. (2014) showed a

TABLE 1. Number of extreme echo cores and DRAs identified in each of the study regions.

Altiplano Amazon Atlantic

Brazilian

Highlands Northern La Plata Southern La Plata

Northern

foothills Sierras de Córdoba

DCCs 346 427 76 535 403 638 97 1019

WCCs 341 2389 1585 1182 898 2315 620 1049

BSRs 22 572 1466 327 103 514 112 44

DRAs 57 150 715 435 610 958 191 789 50 045 123 712 183 614 64 714

TABLE 2. Ratio of the number of extreme echo cores to the total number of DRAs in each of the study regions, expressed as a percentage (%).

Altiplano Amazon Atlantic

Brazilian

Highlands Northern La Plata Southern La Plata

Northern

foothills Sierras de Córdoba

DCCs 0.60 0.06 0.01 0.28 0.80 0.51 0.05 1.56

WCCs 0.59 0.33 0.26 0.62 1.79 1.86 0.34 1.61

BSRs 0.04 0.08 0.24 0.17 0.21 0.42 0.06 0.07

JANUARY 2016 RASMUS SEN ET AL . 359

similar southwestward shift in the occurrence of flooding

events in the spring-to-summer transition associated

with the same storm types examined in this study. Both

flash and slow-rise floods have been associated with

extreme events producing large amounts of rainfall;

thus, this study is consistent with the findings of

Rasmussen et al. (2014). In both the spring and sum-

mer, Figs. 5 and 6 clearly show that the dominant

contribution to the total rain is by storms containing

WCCs in subtropical South America.

Some studies have focused on a narrow maximum of

climatological rain along the Andes foothills in the

northern foothills region (labeled in Fig. 1; Garreaud

1999; Mohr et al. 2014). This feature is evident in Figs. 2

and 3. However, Figs. 5 and 6 show no significant pre-

cipitation along the northern foothills by storms con-

taining extreme radar echoes. This result implies that

the narrow maximum of precipitation along the north-

ern foothills must be due to smaller or weaker non-

extreme echoes that likely form repeatedly in frequent

uplift of warm tropical air by low-level easterly winds

consistently impinging on the tropical Andes foothills. A

full characterization of the storms in the tropical Andes

foothills is beyond the scope of this study but will be

explored in future work.

a. Regional rainfall contributions

Figure 7 shows the accumulated precipitation con-

tributions for each storm type and region in the spring

and summer, expressed as percentages. The designa-

tion DWCC indicates contributions by storms con-

taining both DCCs and WCCs (section 2b). Consistent

with Figs. 5 and 6, the contribution from storms with

extreme characteristics to the climatological pre-

cipitation is very small in tropical compared to sub-

tropical South America (Fig. 7). As previously shown

in Figs. 4b, 4c, 5b, and 5c, the austral spring shows

patterns of robust WCCs in east-central Argentina and

BSRs located farther east along the southwestern edge

of the Brazilian Highlands. The rain contributions in

Fig. 7 echo these results and show the importance of

storms containing BSRs in the austral spring. However,

along the subtropical Andes foothills, the spring and

summer rain contributions are very similar (Fig. 7),

indicating the strong orographic control on storms with

extreme characteristics in the immediate lee of the

Andes during both seasons.

Storms with wide convective cores contribute more

rain than other extreme storm categories in all of the

continental precipitation regions in the austral summer,

consistent with higher ratios of WCC events to DRAs in

Table 2. The variations in contribution by storms with

DCCs, WCCs, and BSRs to the total rain in the sub-

tropics during the austral summer are consistent with the

storm life cycle hypothesis fromRomatschke andHouze

(2010), Matsudo and Salio (2011), and Rasmussen and

Houze (2011) discussed in section 3. Higher percentages

from storms with DCCs in the Altiplano and Sierras de

Córdoba regions along the Andes foothills highlight the

role of terrain features in focusing convective initiation,

as has been recently demonstrated via mesoscale mod-

eling in Rasmussen and Houze (2015, manuscript sub-

mitted to Mon. Wea. Rev.). Farther east over La Plata

basin, storms with DCCs contribute less rain, while the

contributions from storms with WCCs to the total rain

are ;10%–15% higher than in the Sierras de Córdobaregion because of upscale growth and intensification of

convective systems during the austral summer. Moving

farther east where MCSs tend to decay and become

more stratiform in nature, contributions by storms with

DCCs and WCCs decrease over the Atlantic Ocean.

BSR contributions increase from low values near the

foothills out to the Atlantic Ocean.

The southern La Plata basin region shows the highest

contribution from storms with extreme characteristics.

A total of ;43% of the summer rain falls from storms

withWCCs. Given that La Plata basin is the fifth-largest

river basin in the world, having;43%of its warm season

rainfall come from an extreme storm type typically as-

sociated with MCSs demonstrates the considerable role

of storms with WCCs on the hydrologic cycle in sub-

tropical South America. Approximately 95% of the to-

tal summer rain in the southern La Plata basin region is

accounted for by contributions of storms with extreme

convective and stratiform elements (i.e., storms con-

tainingDCCs,WCCs, DWCCs, or BSRs). Table 2 shows

that storms with extreme characteristics in the southern

La Plata basin region make up only ;3% of all raining

events in this region. Thus, ;3% of all events are re-

sponsible for producing;70% of the spring precipitation

TABLE 3. Averaged areas (km2) of DRAs containing extreme echo cores in each of the study regions.

Altiplano Amazon Atlantic

Brazilian

Highlands Northern La Plata Southern La Plata

Northern

foothills Sierras de Córdoba

DCCs 26 766 15 259 54 724 30 117 46 201 48 207 19 865 25 049

WCCs 53 491 42 399 68 497 60 502 70 807 70 411 46 668 40 367

BSRs 96 932 96 914 105 116 108 182 114 625 112 548 98 774 101 126

360 JOURNAL OF HYDROMETEOROLOGY VOLUME 17

and;95%of the summer precipitation in the southern La

Plata basin region. The importance of understanding the

character of storms contributing these very large fractions

of precipitation is an important area for future study.

In contrast to subtropical South America, the more

maritime nature of the precipitation in tropical South

America (Houze et al. 2015) does not promote the

generation of deep or intense convective systems on a

regular basis (Figs. 5, 6). The contribution from storms

with extreme embedded features to the total rainfall in

the tropics is therefore relatively low compared to the

subtropics (Fig. 7). Rain contributions in the northern

foothills and Amazon regions are surprisingly similar

given their differential proximity to the Andes. One

exception is storms containing WCCs in the austral

spring, whose proximity to the Andes affects the mag-

nitude of their rain contribution. However, the influence

of the Andes is much stronger in the subtropics because

of the relationship to convective storm initiation and

development (Romatschke andHouze 2010; Rasmussen

and Houze 2011; Rasmussen et al. 2014; Rasmussen and

Houze 2015, manuscript submitted to Mon. Wea. Rev.).

b. Seasonal variability

Figure 8 shows a monthly time series of the accumu-

lated rain contribution from the three storm types (i.e.,

storms containing DCCs, WCCs, and BSRs) in the

northern foothills and Amazon (Fig. 8a) and the

southern La Plata basin and Sierras de Córdoba regions(Fig. 8b), expressed as percentages. Overall, the rain

contributions from extreme storms in the tropics are

lower, consistent with a maritime-like regime with typ-

ically smaller and less intense convection (Houze et al.

2015). Over the progress of the seasons, the relative rain

contributions from each storm type are comparable,

suggesting that the more maritime nature of convection

continues throughout the spring and summer (Fig. 8a).

The two subtropical regions (Fig. 8b) also exhibit simi-

larities in the storm type rain contribution percentages,

with a notable maximum in contributions by storms

containingDCCs andWCCs in the summermonths (i.e.,

DJF). In contrast, storms containing BSRs are minimum

in summer, likely indicating the presence of fewer

frontal systems to enhance the stratiform echoes in

subtropical South America during that season.

5. Diurnal cycle

To assess the impact of the diurnal cycle on the rain

contributions from storms with extreme characteristics,

time–longitude diagrams representing the diurnal pro-

gression of rain contribution are presented in Fig. 9

during the austral summer. The data are averaged over a

FIG. 5. Geographical distribution of the rainfall contribution to

the total rain by each storm type during the austral spring (SON)

from 1998 to 2012. The contour inside the continent represents the

500-m terrain elevation.

JANUARY 2016 RASMUS SEN ET AL . 361

meridional band bounded by 368–288S, which was cho-

sen because it includes La Plata basin and Sierras de

Córdoba. Storms containing DCCs initiate around 1400

local time near the Andean foothills, which is consistent

with solar heating of the Andes combined with the

SALLJ bringing moisture south from the Amazon to

provide a favorable environment for deep convection

(Velasco and Fritsch 1987; Zipser et al. 2006;

Romatschke and Houze 2010; Rasmussen and Houze

2011; Matsudo and Salio 2011; Rasmussen et al. 2014;

Rasmussen and Houze 2015, manuscript submitted to

Mon. Wea. Rev.). In Fig. 9a, the black contours repre-

sent the rain contribution from storms containing DCCs

only, indicating the strong tendency for such systems to

appear at the time of initiation of convection in the

immediate foothills of the Andes. The eastward-

propagating pattern in Fig. 9a thus indicates how the

DCC and WCC echo characterizations relate to each

other in a diurnal time-sequential sense. The observed

behavior is that storms containing DCCs form at initi-

ation but then grow upscale after initiation to form

storms of mesoscale dimension that contain WCCs as

they propagate eastward. This behavior is consistent

with that hypothesized by Romatschke and Houze

(2010), Matsudo and Salio (2011), and Rasmussen and

Houze (2011).

As storms containing DCCs move eastward, growing

upscale into intense WCCs, they contribute more cli-

matological precipitation over the plains beyond the

foothills (Figs. 9a,b). By late evening to early morning,

the MCSs are more developed and contain WCCs that

FIG. 7. Percentage of the accumulated rainfall contribution from

each storm type (indicated by the colors in the legend) to the total

accumulated precipitation in each region. Values on the left rep-

resent the contribution from the austral spring (SON), while the

ones in parentheses are from the austral summer (DJF).

FIG. 6. As in Fig. 5, but for the austral summer (DJF).

362 JOURNAL OF HYDROMETEOROLOGY VOLUME 17

are nocturnal, cover large areas on average, and tend to

last ;18 h, as suggested by the TRMM observations

(Figs. 9a,b). As the convective elements within the

storms containingWCCs begin to lose their buoyancy in

their mature phase, they transition into stratiform pre-

cipitation. Figure 9c shows that this transition occurs in

the early morning and persists through midday.

However, a sharp cutoff in the precipitation at the edge

of the continent (i.e., ;508W) indicates that the storms

significantly weaken as they move off the coast and over

theAtlanticOcean. Figure 9 echoes the results of studies

investigating the diurnal characteristics and longitudinal

propagation of MCS-related precipitation over the

United States forming near another major mountain

range (i.e., the Rockies) and moving eastward (Trier

et al. 2010).

Rasmussen et al. (2014) recently found that the di-

urnal patterns of extreme-storm-related lightning

exhibit a convective back-building tendency toward the

Andes. Storms containing DCCs only (black contours in

Fig. 9a) show a tendency for the most intense convective

elements to remain linked with the Andes foothills from

the initiation stage through the early morning hours, in

agreement with the back-building hypothesis of

Rasmussen et al. (2014). However, the categories asso-

ciated with larger and more organized convection (color

shading in Figs. 9a–c) show a clear eastward propagation

as MCSs grow, expand, and move eastward during their

diurnal life cycle. Thus, while the convective intensity

along the Andes foothills remains strong from the early

afternoon through the early morning [Fig. 3 in

Rasmussen et al. (2014); Fig. 9a], the precipitation as-

sociated with the eastward-expanding MCSs moves

eastward. A time–longitude diagram representing the

diurnal cycle of precipitation from all summertime

DRA events seen by the TRMM satellite is presented in

Fig. 10. A strikingly similar diurnal cycle of precipitation

is revealed when comparing Figs. 9 and 10, further

supporting the dominant nature of extreme storms and

mesoscale convective systems in the hydroclimate of

subtropical South America.

6. Conclusions

This investigation using 15 years of TRMM PR data

provides insights into the influence of convective sys-

tems with extreme characteristics on the hydrologic cy-

cle in South America. TRMM satellite observations

have enabled an unprecedented characterization of

South American precipitation because most of the

continent is sampled frequently with the same in-

strumentation and because the TRMMPRmeasures the

vertical as well as horizontal structure of the radar

echoes of precipitating convection. To assess the relative

contribution to climatological rainfall by the most ex-

treme configurations over the continent, precipitation

was accumulated from all raining pixels seen by the

TRMM satellite. By identifying those particular types of

extreme radar echo structures that are considered to

approximate the life cycle stages of extreme convective

systems, from newly developed to mature to decaying,

we have been able to map the ratio of extreme-storm-

contributed rain in each of its stages to the continent’s

climatological rain. From a hydrologic and climatologi-

cal viewpoint, this empirical knowledge is useful be-

cause the type of runoff and flooding that may occur

depends on the specific character and life stage of the

convection and precipitation reaching the surface and

FIG. 8. Monthly time series of the accumulated rain contribution, expressed as percentages, from the three storm

types (DCC,WCC, and BSR) in the regions of the (a) northern foothills and Amazon and (b) southern La Plata and

Sierras de Córdoba regions.

JANUARY 2016 RASMUS SEN ET AL . 363

has broad implications for the hydrological cycle in

this region.

An algorithm applied to the TRMM PR data has

identified three distinct categories of storms containing

embedded echoes of extreme horizontal or vertical di-

mension from the TRMM PR data. A significant dif-

ference in how much rain storms containing these forms

of embedded echoes contribute to the climatological

rain in the tropics and subtropics of South America

reveals the different characteristics of the convective

populations in these two climatological regimes. Table 4

shows the relative impact of precipitation from storms

with extreme characteristics in the tropics and sub-

tropics of South America during the austral spring and

summer. Although tropical South America has a higher

number of events and more climatological rain, the

relative impact of total extreme storm precipitation on

the hydrologic cycle of the Amazon is a factor of 2–3

FIG. 9. Time–longitude diagrams representing the diurnal progression of the contribution to

the total rain climatology from (a) DCCs, (b) WCCs, and (c) BSRs for the austral summer

(DJF). The black contour in (a) is the contribution from those events that were classified as

DCC only (see text). The diagrams are averaged over a meridional band bounded by 368–288S.The gray line in (a)–(c) represents the average topographic relief between 368 and 288S with

a max height of 3500m.

364 JOURNAL OF HYDROMETEOROLOGY VOLUME 17

lower than the subtropics. The tropical region of South

America, primarily the Amazon basin, maintains a

generally maritime character with relatively few ex-

treme events yet abundant precipitation. In contrast, the

subtropics, primarily Argentina, have much less rain,

but that rain is explained almost wholly by storms of the

most extreme categories. The subtropical rainstorms are

strongly controlled by the Andes. For these reasons, the

role of extreme storms on the hydrologic cycle is

markedly different between the tropics and subtropics

of South America.

In the subtropical zone of South America, storms

containing deep convective cores contribute only a small

proportion of climatological rain near the foothills be-

cause of their smaller size and typically short-lived and

relatively infrequent nature. Storms containing wide

convective cores contribute a large fraction of the cli-

matological rain in La Plata basin, as they are typically

associated with growing and mature mesoscale convec-

tive systems that are large in area and tend to be long-

lived systems. Storms containing broad stratiform

regions contribute a significant amount of climatological

rain with decaying mesoscale convective systems with

widespread and long-lived stratiform precipitation. In

La Plata basin, storms containing wide convective cores

contribute ;44% of the total warm season rain. The

accumulated contribution from storms containing deep

convective cores, deep and wide convective cores, wide

convective cores, and broad stratiform regions is ;95%

of the total warm season precipitation in La Plata basin.

In La Plata basin, storms with extreme characteristics

only make up ;3% of the total events that occur, so

although these storms are rare, they produce most of the

rain. Since such a large fraction of the total warm season

precipitation is associated with storm types related to

the life cycle of strong MCSs, local populations and the

regional economy critically depend on these types

of storms.

Diurnal time–longitude analysis of the rain contribu-

tion by these three types of storms in the subtropical

zone in summer is consistent with the mesoscale con-

vective system life cycle hypothesis from Romatschke

and Houze (2010), Matsudo and Salio (2011), and

Rasmussen and Houze (2011). Convective initiation in

the late afternoon near the foothills of the Andes

sometimes leads to the upscale growth of convective

echoes to form nocturnal mesoscale convective systems

that begin to expand eastward and then finally decay

into broad stratiform regions farther east. As mentioned

in Rasmussen et al. (2014) and Rasmussen and Houze

(2015, manuscript submitted to Mon. Wea. Rev.), the

most intense convective elements have a back-building

character that tend to remain tied to the terrain from the

late afternoon hours during convective initiation

through the mature stages of the storm in the early

morning hours. This pattern contrasts with the pre-

cipitation associated with larger convection that moves

diurnally in a continuous eastward shift along with the

growing systems.

Convective storms with extreme characteristics have

the greatest impact in La Plata basin in South America.

Given that La Plata basin produces ;70% of the gross

national product for countries within the basin (Mechoso

et al. 2001; World Water Assessment Program 2009),

major rivers in the basin supply;80%of the electricity to

the region through hydroelectric power, and it is the fifth-

largest river basin in the world, an understanding of the

storms that contribute 70%–95% of the rainfall associ-

ated with convective storms in the region is extremely

important. Changes in the timing, frequency, location, or

intensity of the mesoscale convective systems would have

considerable practical implications. In the future, a

greater understanding of the relationship of howmuch of

the hydrologic cycle of SouthAmerica depends on storms

with the specific convective andmesoscale characteristics

described in this paper will provide valuable information

on the hydroclimate of the region. Modeling for fore-

casting of weather and floods is challenged by the variety

FIG. 10. As in Figs. 9a–c, but for the climatological diurnal cycle

of precipitation (mmh21) from all DRAs in subtropical South

America during the austral summer (DJF).

TABLE 4. Climatological rain contribution from DRAs con-

taining extreme echo cores to the climatological rain in the tropical

and subtropical regions of South America during the austral spring

and summer.

Tropics (%) Subtropics (%)

DCCs 1.4/0.5 1.6/1.9

DCCs and WCCs 2.3/0.5 11.2/9.8

WCCs 24.5/13.8 34.5/33.2

BSRs 9.5/10.4 38.4/26.9

Total rain contribution 39.1/25.2 85.7/83.5

JANUARY 2016 RASMUS SEN ET AL . 365

of storm structures and life cycles occurring over different

parts of the continent. As climate changes, these con-

vective behaviors and patterns may change, and accu-

rate modeling will be needed to assess potential

changes. In the present climate, better information is

needed for local water management, high-impact

weather mitigation, and the scientific understanding

of the hydrometeorology of South America with im-

plications for hydrological forecasting and public

safety. This study lays groundwork for such advances.

Acknowledgments. The authors thank Edward Zipser

and two anonymous reviewers for their comments and

suggestions, which have greatly improved this manu-

script. Beth Tully coordinated the graphics. This re-

search was sponsored by NSF Grant AGS-1144105,

NASA Grants NNX13AG71G and NNX10AH70G,

and a NASA Earth and Space Science Graduate Fel-

lowship (NNX11AL65H).

REFERENCES

Awaka, J., T. Iguchi, H. Kumagai, and K. Okamoto, 1997: Rain

type classification algorithm for TRMM Precipitation Radar.

IGARSS ’97: Remote Sensing—A Scientific Vision for Sus-

tainable Development, Vol. 4, IEEE, 1633–1635, doi:10.1109/

IGARSS.1997.608993.

Boccippio, D. J., W. A. Petersen, andD. J. Cecil, 2005: The tropical

convective spectrum. Part I: Archetypal vertical structures.

J. Climate, 18, 2744–2769, doi:10.1175/JCLI3335.1.

Carvalho, L. M. V., C. Jones, and B. Leibmann, 2004: The South

Atlantic ConvergenceZone: Form, persistence, and relationships

with intraseasonal and interannual activity and extreme rainfall.

J. Climate, 17, 88–108, doi:10.1175/1520-0442(2004)017,0088:

TSACZI.2.0.CO;2.

Durkee, J. D., T. L. Mote, and J. M. Shepherd, 2009: The contri-

bution of mesoscale convective complexes to rainfall across

subtropical South America. J. Climate, 22, 4590–4605, doi:10.1175/

2009JCLI2858.1.

Fritsch, J. M., R. J. Kane, and C. R. Chelius, 1986: The contri-

bution of mesoscale convective weather systems to the warm-

season precipitation in the United States. J. Climate Appl.

Meteor., 25, 1333–1345, doi:10.1175/1520-0450(1986)025,1333:

TCOMCW.2.0.CO;2.

Garreaud, R. D., 1999: Multiscale analysis of the summertime pre-

cipitation over the central Andes.Mon.Wea. Rev., 127, 901–921,

doi:10.1175/1520-0493(1999)127,0901:MAOTSP.2.0.CO;2.

Houze, R. A., Jr., 2004: Mesoscale convective systems. Rev. Geo-

phys., 42, RG4003, doi:10.1029/2004RG000150.

——, D. C. Wilton, and B. F. Smull, 2007: Monsoon convection in

theHimalayan region as seenby theTRMMPrecipitationRadar.

Quart. J. Roy. Meteor. Soc., 133, 1389–1411, doi:10.1002/qj.106.

——, K. L. Rasmussen, S. Medina, S. R. Brodzik, and

U. Romatschke, 2011: Anomalous atmospheric events leading

to the summer 2010 floods in Pakistan. Bull. Amer. Meteor.

Soc., 92, 291–298, doi:10.1175/2010BAMS3173.1.

——, ——, M. D. Zuluaga, and S. R. Brodzik, 2015: The variable

nature of convection in the tropics and subtropics: A legacy of

16 years of the Tropical Rainfall Measuring Mission

(TRMM) satellite. Rev. Geophys., 53, 994–1021, doi:10.1002/

2015RG000488.

Huffman, G. J., R. F. Adler, M. M. Morrissey, D. T. Bolvin,

S. Curtis, R. Joyce, B. McGavock, and J. Susskind, 2001:

Global precipitation at one-degree daily resolution from mul-

tisatellite observations. J. Hydrometeor., 2, 36–50, doi:10.1175/

1525-7541(2001)002,0036:GPAODD.2.0.CO;2.

Iguchi, T., R. Meneghini, J. Awaka, T. Kozu, and K. Okamoto,

2000: Rain profiling algorithm for the TRMM Precipitation

Radar data. Adv. Space Res., 25, 973–976, doi:10.1016/

S0273-1177(99)00933-3.

——, T. Kozu, J. Kwiatkowski, R. Meneghini, J. Awaka, and

K. Okamoto, 2009: Uncertainties in the rain profiling algo-

rithm for the TRMM Precipitation Radar. J. Meteor. Soc.

Japan, 87A, 1–30, doi:10.2151/jmsj.87A.1.

Kodama, Y., 1992: Large-scale common features of subtropical

precipitation zones (the Baiu frontal zone, the SPCZ, and the

SACZ). Part I: Characteristics of subtropical frontal zones.

J. Meteor. Soc. Japan, 70, 813–835.

Kummerow,C.,W. Barnes, T. Kozu, J. Shiue, and J. Simpson, 1998:

The Tropical Rainfall Measuring Mission (TRMM) sensor

package. J. Atmos.Oceanic Technol., 15, 809–817, doi:10.1175/

1520-0426(1998)015,0809:TTRMMT.2.0.CO;2.

——, and Coauthors, 2000: The status of the Tropical Rainfall

Measuring Mission (TRMM) after two years in orbit. J. Appl.

Meteor., 39, 1965–1982, doi:10.1175/1520-0450(2001)040,1965:

TSOTTR.2.0.CO;2.

Latrubesse, E. M., and D. Brea, 2009: Floods in Argentina.Natural

Hazards and Human-Exacerbated Disasters in Latin America,

E. Latrubesse, Ed., Developments in Earth Surface Processes,

Vol. 13, Elsevier, 333–349, doi:10.1016/S0928-2025(08)10016-5.

Liu, C., 2011: Rainfall contribution from precipitation systems with

different sizes, intensities, and durations. J. Hydrometeor., 12,

394–412, doi:10.1175/2010JHM1320.1.

Liu, Z., 2015: Comparison of precipitation estimates between

version 7 3-hourly TRMM Multi-Satellite Precipitation

Analysis (TMPA) near-real-time and research products. Atmos.

Res., 153, 119–133, doi:10.1016/j.atmosres.2014.07.032.

Matsudo, C. M., and P. V. Salio, 2011: Severe weather reports and

proximity to deep convection over northern Argentina. At-

mos. Res., 100, 523–537, doi:10.1016/j.atmosres.2010.11.004.

Mechoso, R. C., and Coauthors, 2001: Climatology and hydrology

of the Plata basin. VAMOS Scientific Study Group on the

Plata Basin Doc., 56 pp. [Available online at http://www.

atmos.umd.edu/~berbery/lpb/science_plan.html.]

Mohr, K. I., J. S. Famiglietti, and E. J. Zipser, 1999: The contri-

bution to tropical rainfall with respect to convective system

type, size, and intensity estimated from the 85-GHz ice-

scattering signature. J. Appl. Meteor., 38, 596–606, doi:10.1175/

1520-0450(1999)038,0596:TCTTRW.2.0.CO;2.

——, D. Slayback, and K. Yager, 2014: Characteristics of pre-

cipitation features and annual rainfall during the TRMM era

in the central Andes. J. Climate, 27, 3982–4001, doi:10.1175/

JCLI-D-13-00592.1.

Negri, A. J., T. L. Bell, and L. Xu, 2002: Sampling of the diurnal

cycle of precipitation using TRMM. J. Atmos. Oceanic Tech-

nol., 19, 1333–1344, doi:10.1175/1520-0426(2002)019,1333:

SOTDCO.2.0.CO;2.

Nogués-Paegle, J., and K. C. Mo, 1997: Alternating wet and dry

conditions over South America during summer. Mon. Wea.

Rev., 125, 279–291, doi:10.1175/1520-0493(1997)125,0279:

AWADCO.2.0.CO;2.

366 JOURNAL OF HYDROMETEOROLOGY VOLUME 17

Rasmussen, K. L., and R. A. Houze Jr., 2011: Orogenic convection

in South America as seen by the TRMM satellite. Mon. Wea.

Rev., 139, 2399–2420, doi:10.1175/MWR-D-10-05006.1.

——, S. L. Choi, M. D. Zuluaga, andR.A. Houze Jr., 2013: TRMM

precipitation bias in extreme storms in South America. Geo-

phys. Res. Lett., 40, 3457–3461, doi:10.1002/grl.50651.

——,M. D. Zuluaga, and R. A. Houze Jr., 2014: Severe convection

and lightning in subtropical South America. Geophys. Res.

Lett., 41, 7359–7366, doi:10.1002/2014GL061767.

——, A. J. Hill, V. E. Toma, M. D. Zuluaga, P. J. Webster, and

R.A.Houze, Jr., 2015:Multiscale analysis of three consecutive

years of anomalous flooding in Pakistan.Quart. J. Roy. Meteor.

Soc., 141, 1259–1276, doi:10.1002/qj.2433.

Richey, J. E., L. A. K. Mertes, T. Dunne, R. L. Victoria, B. R.

Forsberg, A. C. N. S. Tancredi, and E. Oliveira, 1989: Sources

and routing of the Amazon River flood wave. Global Bio-

geochem. Cycles, 3, 191–204, doi:10.1029/GB003i003p00191.

Robinson, F. J., S. C. Sherwood, D. Gerstle, C. Liu, and D. J.

Kirshbaum, 2011: Exploring the land–ocean contrast in con-

vective vigor using islands. J. Atmos. Sci., 68, 602–618,

doi:10.1175/2010JAS3558.1.

Romatschke, U., and R. A. Houze Jr., 2010: Extreme summer

convection in South America. J. Climate, 23, 3761–3791,

doi:10.1175/2010JCLI3465.1.

——, and ——, 2011: Characteristics of precipitating convective

systems in the SouthAsianmonsoon. J. Hydrometeor., 12, 3–26,doi:10.1175/2010JHM1289.1.

——, and ——, 2013: Characteristics of precipitating convective

systems accounting for the summer rainfall of tropical and

subtropical South America. J. Hydrometeor., 14, 25–46,

doi:10.1175/JHM-D-12-060.1.

——, S.Medina, andR.A.Houze Jr., 2010: Regional, seasonal, and

diurnal variations of extreme convection in the South Asian

region. J. Climate, 23, 419–439, doi:10.1175/2009JCLI3140.1.Rozante, J. R., D. S. Moreira, L. G. de Goncalves, and D. A.

Vila, 2010: Combining TRMM and surface observations

of precipitation: Technique and validation over South

America. Wea. Forecasting, 25, 885–894, doi:10.1175/

2010WAF2222325.1.

Salio, P., M. Nicolini, and E. J. Zipser, 2007: Mesoscale convective

systems over southeastern South America and their relation-

ship with the South American low-level jet. Mon. Wea. Rev.,

135, 1290–1309, doi:10.1175/MWR3305.1.

Trier, S. B., C. A.Davis, andD.A.Ahijevych, 2010: Environmental

controls on the simulated diurnal cycle of warm-season pre-

cipitation in the continental United States. J. Atmos. Sci., 67,

1066–1090, doi:10.1175/2009JAS3247.1.

Velasco, I., and J. M. Fritsch, 1987: Mesoscale convective com-

plexes in the Americas. J. Geophys. Res., 92, 9591–9613,

doi:10.1029/JD092iD08p09591.

Vera, C., and Coauthors, 2006: The SouthAmerican Low-Level Jet

Experiment. Bull. Amer. Meteor. Soc., 87, 63–77, doi:10.1175/

BAMS-87-1-63.

Wall, C., E. J. Zipser, and C. Liu, 2014: An investigation of the

aerosol indirect effect on convective intensity using satellite

observations. J. Atmos. Sci., 71, 430–447, doi:10.1175/

JAS-D-13-0158.1.

World Water Assessment Program, 2009: Water in a changing

world. United Nations World Water Development Rep. 3,

UNESCO Publishing, 318 pp.

Zipser, E. J., C. Liu, D. J. Cecil, S. W. Nesbitt, and D. P. Yorty, 2006:

Where are themost intense thunderstorms onEarth?Bull.Amer.

Meteor. Soc., 87, 1057–1071, doi:10.1175/BAMS-87-8-1057.

Zuluaga, M. D., and R. A. Houze Jr., 2013: Evolution of the

population of precipitating convective systems over the

equatorial Indian Ocean in active phases of the Madden–

Julian oscillation. J. Atmos. Sci., 70, 2713–2725, doi:10.1175/

JAS-D-12-0311.1.

——, and ——, 2015: Extreme convection of the near-

equatorial Americas, Africa, and adjoining oceans as

seen by TRMM. Mon. Wea. Rev., 143, 298–316, doi:10.1175/

MWR-D-14-00109.1.

JANUARY 2016 RASMUS SEN ET AL . 367